Skillset_Clustering.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Skillset Cluster Analysis\n",
    "\n",
    "The goal in this investigation is to create a dendrogram (for a weight class) to  see if we can observe clustering based on skillset.\n",
    "That is, can we identify those fighters with wrestling, BJJ, Striking backgrounds by applying techniques from machine learning?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# First, run main.py to scrape the necessary data\n",
    "# It may take a while if this is your first scrape.\n",
    "# After your initial scrape, the script will update your database with the latest fights\n",
    "# The script will print the current event being scraped.\n",
    "# !python3 main.py\n",
    "\n",
    "def height_to_inches(h):\n",
    "    if h == '--':\n",
    "        height = 0\n",
    "    else:\n",
    "        h_list = h.split()\n",
    "        height = int(h_list[0].strip('\\''))*12 + int(h_list[1].strip('\\\"'))\n",
    "    return height\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "from os import getcwd, makedirs\n",
    "from os.path import join, exists\n",
    "from datetime import date\n",
    "from scipy.cluster.hierarchy import dendrogram, linkage\n",
    "\n",
    "figure_dir = join(getcwd(),'Figures')\n",
    "if not exists(figure_dir):\n",
    "    makedirs(figure_dir)\n",
    "\n",
    "processed_events_dir = getcwd() + '/UFCStats_Dicts/Processed/'\n",
    "filename = 'All_Fighters.csv'\n",
    "df = pd.read_csv(join(processed_events_dir,filename),parse_dates=['LastFightDate','NextFightDate','DOB'])\n",
    "\n",
    "df = df.drop(df.index[pd.isna(df['LastFightDate'])])\n",
    "df = df.drop(df.index[df.UFCFights < 2])\n",
    "df['stance'] = df['stance'].astype('category')\n",
    "df['height'] = df['height'].apply(height_to_inches)\n",
    "df['weight'] = pd.to_numeric(df['weight'], errors='coerce')\n",
    "df['reach'] = pd.to_numeric(df['reach'], errors='coerce')\n",
    "df['WinRatio'] = df['W']/df['TotalFights']\n",
    "\n",
    "\n",
    "sns.set()\n",
    "\n",
    "SMALL_SIZE = 20\n",
    "MEDIUM_SIZE = 30\n",
    "BIGGER_SIZE = 40\n",
    "TITLE_FONT_SIZE = 35\n",
    "\n",
    "params = {'font.size': MEDIUM_SIZE,\n",
    "          'legend.fontsize': MEDIUM_SIZE,\n",
    "          'legend.title_fontsize': MEDIUM_SIZE,\n",
    "          'figure.figsize': (16, 9),\n",
    "          'axes.labelsize': MEDIUM_SIZE,\n",
    "          'axes.titlesize': TITLE_FONT_SIZE,\n",
    "          'xtick.labelsize': SMALL_SIZE,\n",
    "          'ytick.labelsize': SMALL_SIZE}\n",
    "plt.rcParams.update(params)\n",
    "\n",
    "dpi_save = 200"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The UFC is an evolving sport, so let's focus on the modern era. However, there isn't really a well-defined \"modern era\". The first UFC fight to air on a major network was Cain Velasquez vs. Junior Dos Santos, which was aired live on FOX on Nov. 11, 2011. So let's limit our investigation to 2012-2020. (At the time of composing this document, the last event occured on Nov. 7, 2020.)\n",
    "\n",
    "We need to curate the data to make the comparison fair.\n",
    "The first fight of the card is always 3 rounds, so we need to remove 5-round fights.\n",
    "Also, catch weight is not actually a weight class, so let's remove it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 2168 entries, 0 to 3579\n",
      "Data columns (total 24 columns):\n",
      " #   Column         Non-Null Count  Dtype         \n",
      "---  ------         --------------  -----         \n",
      " 0   FighterName    2168 non-null   object        \n",
      " 1   FighterLink    2168 non-null   object        \n",
      " 2   LastFightDate  2168 non-null   datetime64[ns]\n",
      " 3   NextFightDate  57 non-null     datetime64[ns]\n",
      " 4   W              2168 non-null   int64         \n",
      " 5   L              2168 non-null   int64         \n",
      " 6   D              2168 non-null   int64         \n",
      " 7   NC             2168 non-null   int64         \n",
      " 8   TotalFights    2168 non-null   int64         \n",
      " 9   UFCFights      2168 non-null   int64         \n",
      " 10  height         2168 non-null   int64         \n",
      " 11  weight         2160 non-null   float64       \n",
      " 12  reach          1443 non-null   float64       \n",
      " 13  stance         2057 non-null   category      \n",
      " 14  DOB            2168 non-null   object        \n",
      " 15  SLpM           2168 non-null   float64       \n",
      " 16  StrAcc         2168 non-null   int64         \n",
      " 17  SApM           2168 non-null   float64       \n",
      " 18  StrDef         2168 non-null   int64         \n",
      " 19  TDAvg          2168 non-null   float64       \n",
      " 20  TDAcc          2168 non-null   int64         \n",
      " 21  TDDef          2168 non-null   int64         \n",
      " 22  SubAvg         2168 non-null   float64       \n",
      " 23  WinRatio       2168 non-null   float64       \n",
      "dtypes: category(1), datetime64[ns](2), float64(7), int64(11), object(3)\n",
      "memory usage: 408.8+ KB\n"
     ]
    }
   ],
   "source": [
    "# df.head()\n",
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df_modern = df[df.LastFightDate >= pd.Timestamp(2019,11,22)].copy()\n",
    "# df_modern = df_modern[(df_modern.weight > 156) & (df_modern.weight<=171)]\n",
    "# df_modern = df_modern[(df_modern.weight > 206)]\n",
    "# df_modern = df_modern[(df_modern.weight > 136) & (df_modern.weight<=146)]\n",
    "df_modern = df_modern[(df_modern.weight > 186) & (df_modern.weight<=206)]\n",
    "# Z = linkage(df_modern[['TotalFights','UFCFights','SLpM','StrAcc','SApM','StrDef','TDAvg','TDAcc','TDDef','SubAvg']], 'ward')\n",
    "# Z = linkage(df_modern[['SLpM','StrAcc','SApM','StrDef','TDAvg','TDAcc','TDDef','SubAvg']], 'ward')\n",
    "Z = linkage(df_modern[['TotalFights','UFCFights','WinRatio','SLpM','SApM','TDAvg','TDDef','SubAvg']], 'centroid')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1440x2160 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig = plt.figure(figsize=(20, 30))\n",
    "dn = dendrogram(Z, labels=df_modern['FighterName'].tolist(),\n",
    "                orientation='right',\n",
    "                leaf_rotation=0, leaf_font_size=6)\n",
    "plt.savefig(join(figure_dir,'dendrogram_test.png'), dpi=dpi_save)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# Select fights from 2012 and forward\n",
    "df_modern = df[df.EventDate >= pd.Timestamp(2012,1,1)].copy()\n",
    "# Ignore 5 round fights\n",
    "df_modern = df_modern[df_modern.RoundFormat.str[0] == '3']\n",
    "# Remove catch weight bouts, as they are not a standard weight class\n",
    "df_modern = df_modern[df_modern.WeightClass != 'Catch Weight']\n",
    "\n",
    "df_modern['Year'] = df_modern.loc[:,'EventDate'].dt.to_period('Y')\n",
    "# Need to identify which outcomes count as finishes\n",
    "method_map = {'Submission': True, 'KO/TKO': True, 'TKO - Doctor\\'s Stoppage': True, 'DQ': True,\n",
    "              'Could Not Continue': True, 'Decision - Split': False, 'Decision - Majority': False,\n",
    "              'Decision - Unanimous': False, 'Overturned': False\n",
    "             }\n",
    "df_modern['Finish'] = df_modern.loc[:,'Method'].map(method_map)\n",
    "\n",
    "\n",
    "# We want to include overturned fight outcomes, because a daily fantasy sports tournament will not\n",
    "# be affected by a fight outcome that was overturned weeks later.\n",
    "# Overturned bouts do not list the method of finish.\n",
    "# However, a finish can be inferred from round and round time of the fight.\n",
    "df_modern.loc[(df_modern.Method == 'Overturned') & \\\n",
    "              ~((df_modern.Round == 3) & (df_modern.RoundTime == '5:00')),'Finish'] = True\n",
    "\n",
    "# Determine first fight on cards\n",
    "# df_modern['TotalFights'] = df_modern.CardPosition.str.split(' of ').str[1].astype(int)\n",
    "# df_modern['CardPosition'] = df_modern.CardPosition.str.split(' of ').str[0].astype(int)\n",
    "# df_modern['FirstFight'] = df_modern['CardPosition'] == 1\n",
    "\n",
    "# Order the weight classes by weight\n",
    "cats =['Strawweight','Flyweight', 'Bantamweight',\n",
    "       'Featherweight', 'Lightweight','Welterweight', 'Middleweight',\n",
    "       'Light Heavyweight','Heavyweight']\n",
    "df_modern['WeightClass'] = df_modern.WeightClass.astype('category')\n",
    "\n",
    "df_modern['WeightClass'] = df_modern['WeightClass'].cat.reorder_categories(cats, ordered=True)\n",
    "\n",
    "\n",
    "total_events = len(df_modern.EventName.unique())\n",
    "date_today = date.today().strftime(\"%B %d, %Y\")\n",
    "\n",
    "\n",
    "print('From January 1, 2012 to {}, there have been {} UFC events.'.format(date_today, total_events))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "pycharm": {
     "name": "#%% md\n"
    }
   },
   "source": [
    "What are indicators of skillset?\n",
    "\n",
    "Grapplers should have more takedowns than strikers, maybe takedown defense\n",
    "Wrestlers tend to secure position after a takedown, while BJJ'ers advance?\n",
    "Are BJJ'ers more likely to reverse? More Subs/Sub attempts?\n",
    "\n",
    "Can the clustering seperate well-rounded fighters from those who are one-trick ponies? Will champions be clustered?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df_modern[''].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "overall_rates = df_modern.groupby(['FirstFight']).Finish.mean()\n",
    "year_ff = (df_modern.groupby(['Year','FirstFight']).Finish.mean()*100)\n",
    "ax = year_ff.unstack().plot(kind='barh',figsize=(16,9), color=['red','blue'])\n",
    "_ = ax.set_title('UFC Finish Rate by Year: First Fight vs. All Others')\n",
    "_ = ax.set_xlabel('Finish Rate (%)')\n",
    "total_finish_rate_title = 'Finish Rate\\n2012-2020'\n",
    "total_finish_rate_true = '    {}%    '.format(round(overall_rates[True]*100,1))\n",
    "total_finish_rate_false = '    {}%    '.format(round(overall_rates[False]*100,1))\n",
    "\n",
    "\n",
    "x_box = 81\n",
    "y_box = 2.\n",
    "_ = ax.text(x_box, y_box, total_finish_rate_title, style='italic', color='black', fontsize=25,\n",
    "        bbox={'facecolor': 'white', 'alpha': 0.85, 'pad': 10})\n",
    "_ = ax.text(x_box, y_box - 0.75, total_finish_rate_true, style='italic', color='white', fontsize=25,\n",
    "        bbox={'facecolor': 'blue', 'alpha': 0.85, 'pad': 10})\n",
    "_ = ax.text(x_box, y_box - 1.5, total_finish_rate_false, style='italic', color='white', fontsize=25,\n",
    "        bbox={'facecolor': 'red', 'alpha': 0.85, 'pad': 10})\n",
    "_ = ax.set_xlim(right=100)\n",
    "\n",
    "handles,labels = ax.get_legend_handles_labels()\n",
    "handles = [handles[1], handles[0], handles[1]]\n",
    "labels = [labels[1], labels[0]]\n",
    "_ = ax.legend(handles,labels,title='First Fight\\n  on Card')\n",
    "\n",
    "plt.savefig(join(figure_dir,'Finish_Rate_First_Fight_vs_All_Others.png'), dpi=dpi_save)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see 6 of the past 9 years have had higher finish rates in the first fight of the event than the rest of the fights. Furthermore, 3 of the last 4 years have this property. (It's interesting to note that both finish rates for 2019 are the lowest finishes rates from 2012-2020. Guess it wasn't a great year for fans.)\n",
    "\n",
    "Let's see how the finish rates compare by weight class."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "wc_ff = df_modern.groupby(['WeightClass','FirstFight']).Finish.mean()*100\n",
    "ax = wc_ff.unstack().plot(kind='barh',figsize=(16,9),color=['red','blue'])\n",
    "\n",
    "first_fight_middle_rate = round(wc_ff.loc[('Middleweight',True)],1)\n",
    "first_fight_middle_count = sum(df_modern[df_modern['FirstFight']]['WeightClass'] == 'Middleweight')\n",
    "\n",
    "_ = ax.set_title('UFC Finish Rate by Weight Class (2012-2020)')\n",
    "\n",
    "_ = ax.set_xlabel('Finish Rate (%)')\n",
    "_ = ax.set_ylabel('Weight Class')\n",
    "handles,labels = ax.get_legend_handles_labels()\n",
    "handles = [handles[1], handles[0], handles[1]]\n",
    "labels = [labels[1], labels[0]]\n",
    "_ = ax.legend(handles,labels,title='First Fight\\n  on Card')\n",
    "\n",
    "plt.gcf().subplots_adjust(left=0.2)\n",
    "plt.savefig(join(figure_dir,'Finish_Rate_by_Weight_Class.png'), dpi=dpi_save)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "print('Every weight class, with the exception of Featherweight (145 lbs.), has a higher finish rate \\\n",
    "for the first fight of the night!\\nThe middleweight division has a finish rate of {}% when it\\'s the \\\n",
    "first fight on the card!\\nThe high finish rate of middleweights is calculated from {} fights.'.format(first_fight_middle_rate, first_fight_middle_count))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Notice that the finish rates for fights that are not the first on the card increase as the weight classes get heavier (115 lbs. at Strawweight up to a maximum weight of 265 lbs. at Heavyweight). This is to be expected, as heavier fighters tend to be more dangerous.\n",
    "\n",
    "While we're at it, let's see if the COVID pandemic has affected the finish rate. March 14, 2020 (Pi Day!) was the last UFC fight before the schedule was severely disrupted."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# POST-COVID\n",
    "df_modern['PostCOVID'] = np.where(df_modern.EventDate > pd.Timestamp(2020,2,14), 'Post-COVID', 'Pre-COVID')\n",
    "\n",
    "ax = (df_modern.groupby(['PostCOVID','FirstFight']).Finish.mean()*100).unstack().plot(kind='barh',figsize=(16,9),\n",
    "                                                                            color=['red','blue'])\n",
    "_ = ax.set_title('Effect of Pandemic on UFC Finish Rate')\n",
    "_ = ax.set_xlabel('Finish Rate (%)')\n",
    "_ = ax.set_ylabel('')\n",
    "_ = ax.set_xlim(right=70)\n",
    "\n",
    "handles,labels = ax.get_legend_handles_labels()\n",
    "handles = [handles[1], handles[0], handles[1]]\n",
    "labels = [labels[1], labels[0]]\n",
    "_ = ax.legend(handles,labels,title='First Fight\\n  on Card')\n",
    "\n",
    "plt.savefig(join(figure_dir,'Effect_of_Pandemic_on_UFC_Finish_Rate.png'), dpi=dpi_save)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Seems the pandemic era of UFC as resulted in higher finish rates than previous years (still 2012-2020). The UFC initially had trouble getting fighters into the USA, which likely led to more fighters with less experience getting into the premier MMA organization. This trouble led to the creation of fight island in Abu Dhabi.\n",
    "\n",
    "Certainly interesting, and needs further investigation, but let's get back to the hypothesis. Can the higher finish rate for the first fight be correlated to a mismatch of skill level? To investigate, we will look at the empirical cumulative distribution function (ECDF) of the total UFC fights of the fighters in the first fight at the time of their fight. If I've lost you at this stage, have no fear, I will interpret the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "def ecdf(data):\n",
    "    \"\"\" Compute ECDF \"\"\"\n",
    "    x = np.sort(data)\n",
    "    n = x.size\n",
    "    y = np.arange(1, n+1) / n\n",
    "    return(x,y)\n",
    "\n",
    "def count_prev_fights(name, date):\n",
    "    # search all events here for fighter's name and sum the fight experience they have prior to the considered bout.\n",
    "    matches = df.apply(lambda row: ((row['FighterName1'] == name) | (row['FighterName2'] == name)), axis=1)\n",
    "    return sum([(df.loc[i,'EventDate'] < date) for i, x in enumerate(matches) if x])\n",
    "\n",
    "# This part will be very slow if we perform the operation for every fight.\n",
    "# We only care about the first fights, so let's take what we need from the DataFrame\n",
    "df_ff = df_modern[df_modern['FirstFight']].copy()\n",
    "df_ff['PrevFights1'] = df_ff.apply(lambda row: count_prev_fights(row['FighterName1'],row['EventDate']), axis=1)\n",
    "df_ff['PrevFights2'] = df_ff.apply(lambda row: count_prev_fights(row['FighterName2'],row['EventDate']), axis=1)\n",
    "df_ff['PrevFightSum'] = df_ff['PrevFights2'] + df_ff['PrevFights1']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "df_finish = df_ff[df_ff['Finish']]['PrevFightSum']\n",
    "df_no_finish = df_ff[~df_ff['Finish']]['PrevFightSum']\n",
    "x_fin,y_fin = ecdf(df_finish)\n",
    "x_no,y_no = ecdf(df_no_finish)\n",
    "# data offset by 0.2 for visability\n",
    "plt.scatter(x=x_no+0.2, y=y_no*100,label='No Finish',alpha=0.5,marker='d',color='red',s=100)\n",
    "plt.scatter(x=x_fin, y=y_fin*100,label='Finish',alpha=0.5,marker='*',color='blue',s=100)\n",
    "_ = plt.title('Empirical Cumultive Distribution Functions')\n",
    "_ = plt.xlabel('Total UFC Fights Prior to Bout')\n",
    "_ = plt.ylabel('% of Bouts with = or < Experience')\n",
    "lgnd = plt.legend()\n",
    "lgnd.legendHandles[0]._sizes = [500]\n",
    "lgnd.legendHandles[1]._sizes = [500]\n",
    "\n",
    "plt.savefig(join(figure_dir,'Empirical_Cumultive_Distribution_Functions.png'), dpi=dpi_save)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# Determine the total number of fights with the greatest difference in ecdfs\n",
    "greatest_diff = -1\n",
    "total_fights = -1\n",
    "for i in range(10):\n",
    "    tot_fights = i\n",
    "    diff_of_max = max(y_fin[x_fin<=tot_fights]) - max(y_no[x_no<=tot_fights])\n",
    "    if diff_of_max > greatest_diff:\n",
    "        greatest_diff = diff_of_max\n",
    "        total_fights = i\n",
    "print('Interpretation: Of the fights that were first on the card, {}% of matchups with a finish \\\n",
    "had a combined UFC fight history of 1 fight or less. (Fight history previous to fight considered.) \\\n",
    "For the fights that went to decision, {}% of matchups had a combined UFC fight history of 1 fight or less.\\n\\n\\\n",
    "We see the greatest difference in the ECDF for {} fights or less. {}% of matchups with a finish \\\n",
    "had a combined UFC fight history of {} fight or less.\\\n",
    "For the fights that went to decision, {}% of matchups had a combined UFC fight history of {} fight or less.\\n\\n'.format(\n",
    "    round(max(y_fin[x_fin<=1])*100),\n",
    "    round(max(y_no[x_no<=1])*100),\n",
    "    total_fights,\n",
    "    round(max(y_fin[x_fin<=total_fights])*100),\n",
    "    total_fights,\n",
    "    round(max(y_no[x_no<=total_fights])*100),\n",
    "    total_fights))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "These differences in ECDF are not as extreme as I had hoped, but they are present, nonetheless. We need to investigate further.\n",
    "\n",
    "Before moving on, you may wonder why there are fighters with so much experience fighting in the first fight of the night. Generally, when there is a big main event fight, the UFC likes to stack the card. Or, an experienced fighter is on a losing streak. For example, UFC 249: Ferguson vs. Gaethje had 'Smiling' Sam Alvey fighting first. Alvey had 19 UFC fights entering that bout, but was on a 4-fight losing streak. (I hear Sam is still smiling, though.)\n",
    "\n",
    "So, we've found some differences in the data based on whether or not the fight ended in a finish. The question remains: Is the observed difference statistically significant? Let's check the p-value. (Don't worry, I'll interpret.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "# Null hypothesis significance testing\n",
    "# Statistical significance?\n",
    "def perm_fights(df_1,df_2,tot_fts):\n",
    "    all_first_fights = np.concatenate((df_1,df_2))\n",
    "    all_first_fights = np.random.permutation(all_first_fights)\n",
    "    df_finish_perm_sample = all_first_fights[:len(df_1)]\n",
    "    df_no_finish_perm_sample = all_first_fights[len(df_1):]\n",
    "    x_fin,y_fin = ecdf(df_finish_perm_sample)\n",
    "    x_no,y_no = ecdf(df_no_finish_perm_sample)\n",
    "    return max(y_fin[x_fin<=tot_fts]) - max(y_no[x_no<=tot_fts])\n",
    "\n",
    "test_count = 10**4\n",
    "diff_of_max_perm_replicates = np.empty(test_count)\n",
    "for i in range(test_count):\n",
    "    diff_of_max_perm_replicates[i] = perm_fights(df_finish,df_no_finish,total_fights)\n",
    "\n",
    "# _ = plt.hist(diff_of_max_perm_replicates,density=True,bins=25)\n",
    "sns.histplot(diff_of_max_perm_replicates, stat='probability', bins=30, color='blue')\n",
    "_ = plt.axvline(greatest_diff, color='red', linestyle='dashed', linewidth=3)\n",
    "plt.title('Testing The Null Hypothesis')\n",
    "plt.xlabel('Difference in ECDFs for 3 fights or less')\n",
    "\n",
    "plt.savefig(join(figure_dir,'Testing_Null_Hypothesis.png'), dpi=dpi_save)\n",
    "\n",
    "p = np.sum(diff_of_max_perm_replicates >= greatest_diff) / len(diff_of_max_perm_replicates)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "print('We get p-value = {}.'.format(p))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For a p-value to be statistically significant, it needs to be less than 0.05. Therefore, the analysis does not back up the hypothesis that the first fight on the card is more likely to have a finish due to a mismatch in skill level.\n",
    "\n",
    "If you've made it this far into the analysis, you may wonder why I bothered to highlight my failed hypothesis. It is important to come to terms with a failed hypothesis. While I was an undergraduate at the University at Buffalo, a graduate student in physics gave me an anecdote that has resonated throughout the years: Earning a PhD only takes about 5 days of good research, the rest of your time will be spent dealing with your failures. Perhaps a slight exaggeration, it's still a great reminder that there is always something to learn from your mistaken hypotheses. I leave you with a quote from Sir Anthony Hopkins' character in West World:\n",
    "\n",
    "> \"Mistakes\" is the word you're too embarrassed to use. You ought not to be. You're a product of a trillion of them. Evolution forged the entirety of sentient life on this planet using only one tool: the mistake.\n",
    "\n",
    "Let me know if you have questions, comments, or criticisms. (Especially criticism, as I aim to learn more.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "pycharm": {
     "name": "#%%\n"
    }
   },
   "outputs": [],
   "source": [
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}